37 and representation or used denotes

a which floating
mantissa to be in point It theis different
digits 128
form therefore
numbersin refers
can a, 16. accurate M=
numbers x denotes
as or
binary
the e digit
number integer of
binary B written exponent10 number =27,
if the normalized for
of So then leading 8, more
1100;
binary system
The an as the2, inequation.n different
n
to numbers.
real
B= integer B=2,
be is to 0, as is, maximum
(1AB.9C),1e numbers. a_1#
referred the
C= 1100), A can e
and that the to as n
three binary have
16.
thatthat accordingpositive
referred larger are
and decimal representation if following
of or B Note
particular e)
1011.1001 111), B base is (the we
1001 blocks thehexadecimal10 _g...a_)x a_,
factso that expone
C)16 number (110101011.100111),
= : form. part, The 7094
1100), 100 or withinequality the is point n
=
use the .a_ja_... mantissa the evident(base),
make 4 8 -M.
9 011. computersareor system In normalized
to mantissa satisfies IBM
0110 (19B.D6C)1A. 1011,
hexadecimal1010 representation 2 exponent.refers
hexadecimal the
we (653.47)8
=
. = pointa_2 =
or calculators
3 B
6 101 B the B-1.
s
0sa, m is of of
quantities
=(.1101 = form numbers assume the
(.a_j viz. the case
(1AB.9C)16=(0001
(110
5
floating satisfy point
the
M.
mse
exponentand it value
(.D
= B and
1010, octal (6 digital x=#of
bracket
thethe in
in point integers in
ERRORS = the = = digits floating or mantissa largest
(110011011.110101101l),
(.1101011011), the
(110101011.100111),octal pocket
we a_; as as point point the
insta
Convert= electronic its
system the theto known The the
A For digits referred
in the that
AND 0001, pointtheor calculatorare inside and decimal
nonzero.that also
the notewort
(the
form. calculators The is For
form. viz. number float
SYSTEMS 1.
the
..,a-n representation
a#0 note are in M com
binary is
guantity mantiss
EXAMPLE
octall Floatingit
Generally B. e to or M digits
that of of integer as is to
point
variant a a_2 mantissa and x.
pretinent
a or power that the
number
the Note the deskof a_1, so of is
NUMBER computer The adjusted octal m number
is base the fact It of
. to and
get the
where where the
which 1.5 some the point ypes
then and the theor
in the is is in
 in left the
memor no
we. in are the
machine
adouble form put number the to factor in
n. It importance
numbers. allows We ensures
series as binary
SYSTEMSorintegerprecision. bits be to the the 2-2 aresulting floating
type point
Therefore form.
length with precisIon. given can i.e., in
important Then error to
normailized k binary -k)2 from binary
Burrough the which left. numbers of stored later
data which
thedouble
unless
precision these the
more is ... representation
understood fractions.
NUMBERtheon double form as
the the 4-2 in Thus numeric number
the
number accurately)
that other single of said l00.01 very is
i.e.
the binary of
the depends of normalized multiplying
to (0.a_j which (.0011),.
in aprogramme are representation full bits binary thedevice normalised
for note 2 a converts
the
then
n=14 over with is is mantissa well
numbers point a_) x in 4.25 three of (which point 13 64 as more
is mantissa,
in
the
computing
andto and programme
representation be 4-bit= same
pertinent This
of 2; point case. a a' which process
64 point ... x.
floating
will number stores
by 3 1
cases
in
length decimal floating i>2. representation digit
x the 64 exactly
= normalised I=t(a-1@- number 0.10001 this of computer given machine
computing
the 101.0011
numerical
somne
M precision, floating binary machines). 16
6. is word for in significant normalised a which of
is
the feed is 13 4-bits
= It and 1 the point a which in
n of
in or as the1 is a we machine
to
=
(.001101)2(.0011),
64
(and number
16. on. the single of
execution numbers
point
the
;=0of binary form whichfloating over corresponding stores a
execution numbers float) by Suppose to exactly
B= so as single in representation of done the
of
x a point computations equal then x
0.1101 binary
result
360andregarded floating number a_j=land (orbit number
of
acomputing
in
a a.
within of binary the calculations use
IBM64 (say) normalised
non-zero importance 12. and data
time of shift
floating is as finalthe
operates timne example device. we formis a
= maximum EXAMPLE is a
stored
of M 6 the thethe Binary binary to for if our
number the Write
a storing
case 1s
13, = about point
Where as numerical a` Instead,
point sotre
machinen thattheSince normalized
leading digital
The 0.00l101 in
n= For The numerical example. storing EX.
in machine
If twice A floating
so which floatingin to accuracy
B=8,
while upon.
noted belowThe 23
by a the error
the out a able
38 1s of in to a in to
a
39 ... integers
15.1 andfirstof the storing 12345.67 k-digit ...(1.5.2)
computing are decimal
digits is uNhg consIder
u
(1.5.1) There x place
mantissa
(the number, if
the are torm 011
form the floating kth
Thus decimal
ott
rounded to
positive respectively.
a_j representation
number. numbers digit.
off mantissa
0
point
the
ensures the f(x). place lAus. necessary
a chop
normalised its that of of correspondingdecimal 3rd
are memory at tloating
decinmal
as representation mantissa
simply by the is
fact that
defined the after
mantissa
a9... This denoted otlen
the representation,
10 in
Theof kth. off 3rdnormalised
digits
form.
representation the x to article. 3, chopped is
10
X a_1, is to 0.123 the the 10. = the
in x refers (i). in the 10° is method is k the
the it
and normalisd
decimal number and form at off nextx a_X for if if at operation
numbercondition form X of
number h...
a_p x round Then 4-digit
zero) Normalised of form a-...
a...
..., decimal
10
point
Cne the
normalizedmantissa x0.123l02
10 x0.124
3, the thedecimal in aof
normalised >x to
in point =t0.a_1
10. numbers
ns or mantissas 0.1234567 x=#0.a_j termination.
is in arithmeti
i=2, of digits rounding 201.1203
decimal
written negative satisfies method ×
floating x=0.12364
()
1sa_1S9,
...
s9; a.o exponent. normalised significant floating
normalised the
=
fl(x)
1.0. the terminating fl(¢)
ERRORS a
numberof
be (positive, a_; 0.a_1 as other normalised various
0s point) in andstore in
represented this about then Approxim
a can that x
normalised
number Let
ANDof r (ii) ± part
the
as
0.1
devices In the
performingdiscuss
of (1.5.1), 14. out
Represenstation
number calleddecimalnoted between 13.
number by and the
integer known carrying
SYSTEMS fractional EXAMPLE obtained k-digit EXAMPLE Write
be digital are realby shall by
is -(k+2):
a-(k+1)»
real theis to lies given of represented
an e maximum 0.00123 a EX,chopping.
integer x is For ways We The In
Any
NUMBIER is that
such The after number mantissa
All machine. k-digit is
e It is which place. 1.6
where and form two
thedigit of
a
 arises
when even.place.
as becomes
becomes 3rd
places number.as such
machine oft
technigue defined
are of 0.01).
(not
b: wavg
differscalled to
the known round that roundto
means
is referred
k-decimal 3-decimal to
AND Computations
storage following desire while it placesin
places increase significant
3and
sometimes if given computing no
mainly usually
decimal
afinite numeric digit is However For
SYSTEMS such is 0.014
number
limited we place are 3-decimal the which there
1,0, off.
the
3-decimal
upto
the Suppose as to to
mav simply become
round
This which
situation (iii) 2,
or of
decimal it off off
and computations
a
approximation computation.
Unfortunately, are
a
their
NUMBER storesnemory (
leaverounded
and by errors of digits
to 5 l of
word-length
digits
either 2.103 rule8 digits
numbers. to rounding
exactly done right
ofdigital kth and (iü) off off certain or
because
word off
rounding
of in round
operations number a the
figures significa
the odd be rules rounded of its
number
done let is 3.42543.
in
places.
decimal one. to after an process to
erroneous. following
each off and at is place error tollowing
computers done saidwhile becomes the digits the k-significant
all
the nt
on is round
are changed.
by a_ after 1.,2435
and x when decimal to introduces làarge,
arithmetic two
numbcr
ofdepending
fnite
This numberbe increased
if is une number.
of
words one down 3.42543
x approximation 3.
digits
digit
to
to
followed hignly by
a retained.
to
is not by to rounding
number
40.2316
4th
off qunte auring
cases
and off significant
of
numbers. numbers, a alteration as approximation rounded andthe 3) to rounded
of is is number
and number
of of
numbers are1s
numbor consists a_, a_; increase the errorsome
place the number result
numbers)
digit
of
over are thenthen
number in is digits kthwhen
of following
be The Further, an
digit of in digits
Significant(between
non-zero 1
pointoff The following
rules
decimal it inal the
carried approximation
2.by
chopping. <5, >5,ifa_k41)=5,
to
15. odd).
is
place
afteris of the growth
reduced significant
example of the
0.0136
machine. rounding
machine
to resulting
said the the 3.425
instance,
For off of the off
decimal ak4) EXAMPLE If at
Rounding
are
ahal) (ii)).
(i)): (sinceDumber off
Rounding
offerror. the first 1,0 number
kth the is that
with COmputer rounding
up. roundingg be numbe
theForare
Then (i) (rule
(rule
by upto if if The mniter
ont can
The
(Z) (2) (Zz) rule 13.244
decimal
numbers n ter
the1. A 0.0103the
digits
40.232
3.425 errorg be the
sm (hy off
ffnit The to
The is An
40
 ERRORS
SYSTEMS AND 41
NUMBER
number, we discard all the digits after
Inchopping of a chopping the number 3.012466 at the certain decimal place in
instance, 4th decimal place we
number. For which is the result of
the
3.0124
is the (notof3.0125)
result
rounding off of the decimal3.0125
rounding off of the number 3.012466 at the 4thnumber place).which
get

1.7 k-digit arithmetic

Generally computations are performed by high speed electronic digital machines.
Such machines while doing arithmetic operations (such as addition,
subtraction,
multiplication, dividision) on numbers, first of all transform them intotheir corresponding
k-digit normalised floating point form (k-being fixed for a particular machine). This is
known as k-digit arithmetic done by a computing machine. To be more clear, consider
the addition of the numbers
by a computer for which k=
k 4 (say)
r= 0.1234501 x 10° and y=0.132045 x 103
then
If chopping is done by the machine,
fl(x) = 0.1234 x 10°, fl(y) = 0.1320 x 10
The result of addition by the machine which is capable of doing 4-digit arithmetic is
fl(x) +fly) =0.2554 x 10
Therefore the user of the machine gets 0.2554 x 10° as the output whereas the
actual sum is 0.2554951 x 10; hence an error equal to 0.0000951 x 10 =.951 x 10 is
committed.

EXAMPLE 16. Using 5-digits floating point arithmetic, find the product of
1
; and y= 5
Let denote the multiplication-operation in 5-digit floating point arithmetic.
x==0.33333...
3

fle) =0.33333; (chopping being used)

5 =0.7142857
y=7
fly)=0.71428. (chopping being used)
Now
fle) xflo)
=0.33333 x0.71428
= 0.2380909.
fU(«) xfy) = 0.23809; (chopping being used)
t*y=0.23809.
Xand y.This result agrees upto 5-decimal placos with the exact value of the product of
EX. Find the product of 0.2431 x10 nnd 0.491 0 *ing 4-digit arithmetic.
|Ans. 0.1024 x 1061
 vale 4ctual
tosung
series nfnite he
uee todue arite
d chapped are cugt Choppg
entra dTho
Rrutrt
aS6YY A56428
deng
ge todueoCLurto nunber he
iaxe cduad
by and dala betor
becen erereor af uye Such
calculetors
tamathomotro
ble Cucha aideconpuhng he
teg data approunmate to due eihor
n enhe enic errcoe Jaherest
Erva: 0Tahenent
2otutonoparornat1on ertto
o ondsolutun
eturesn
te disercano numacal he
 uctn
,uctso)
W=
hratbn meay Caueay
rauoceh altod
Cased Erecortrtauncaton Local
The ercore percats
be
Aboluto
Erertor Absoluto
al ho upto yale unatd
Paga:
Date
0s01 2390

ebed